And it's off-loading to the training to simulation etc which causing much of the problem - as pointed out by by Bill Sweetman (the man you love to hate) in AW&ST in their analysis of the report and the JO response....
Opinion: F-35 Software Fixes Likely To Take Time
There are two levels of concern about the latest critical memo on the F-35 Joint Strike Fighter (JSF) from the Pentagon’s director of operational test and engineering (DOT&E), Michael Gilmore.
The first is that it is time for the Defense Department to resolve the friction between the F-35 Joint Program Office (JPO) and DOT&E, which has now reached the status of public flaming. The second is the question of whether late and defective software is a feature, rather than a bug, of any defense system as complex as the JSF........
Delays should not entirely be blamed on management, because what emerges from a review of Gilmore’s reports is the multidimensional complexity of the JSF software challenge.
Essential JSF software runs on at least four platforms: the airplane; the Autonomic Logistics Information System (ALIS), which is supposed to manage maintenance and support; reprogramming laboratories, which develop the mission data file (MDF) software that allows fused sensors to identify threats and targets; and simulators used for training and mission rehearsal. All are functionally interlinked: ALIS and training devices need to reflect the configuration of the aircraft hardware and software on the ramp. Simulators and the airplane run on updated MDFs. The latter connection becomes crucial as air forces move toward live, virtual, constructive training, where pilots fly against synthetic threats and targets, because the MDFs determine how those objects are detected and displayed. All have suffered delays, according to Gilmore’s reports.
There is no end-state to development. The goal is to update the software on a two-year cycle, so there will usually be two standards in service at any one time in the U.S., plus customized MDFs for export customers. Much of the software must be validated to a life-or-death level. Aircraft-borne code changes will require some level of regression testing (to ensure they don’t disrupt flight-critical functions). The intelligence that allows the system to distinguish a missile launcher from the village market bus resides in the MDFs. Under this pressure, the temptation is to do exactly what Gilmore says is happening: patch the problems and build each release on top of the next. But as the coders say:
Ninety-nine little bugs in the code,
Take one down, patch it around,
117 little bugs in the code.
Will this situation improve? Possibly. The JPO has been working since 2011 to meet each customer’s demand for three initial operational capability standards—2B for the Marines, 3i for the Air Force and the definitive 3F—that are being developed on overlapping schedules. If that hadn’t been done, the production effort would have had to slow down (nobody wanted to see hundreds of non-operational jets sitting on ramps at the end of 2017), which would have cost a lot of money. The end of the multitracking will be a relief.
Unfortunately, the overarching customer is not done. Shocked by the cost of upgrading highly integrated custom software on the F-22 and other programs, the Pentagon wants to require open systems architecture (OSA) in all new programs and retrofit it to the F-22 and F-35, to bring more competition and commonality into upgrade efforts. “Open mission systems are key to everything we do,” Air Combat Command leader Gen. Herbert Carlisle said late last year. William LaPlante, the Air Force acquisition chief until December 2015, wanted to go to OSA during the F-35’s Block 4 upgrade process—but it has to compete for time and money with badly needed new weapons and fixes left over from Block 3F.
Can the F-22 and F-35 migrate to OSA and reach the sunlit uplands of regular, on-time upgrades on a stable basis? Probably, but it will take more time and money than people think. The lesson is that what we want software to do in our weapons, and when we want it, has to be scrutinized, Red-teamed and matched to assets as rigorously as any other aspect of system requirements.